December 16, 2025
Stars up, pitchforks out
7 Years, 2 Rebuilds, 40K+ Stars: Milvus Recap and Roadmap
40K stars, fresh features, and a 'Just use Postgres?' brawl
TLDR: Milvus hit 40k GitHub stars and rolled out native hybrid search that mixes keyword and meaning-based results. Comments split between fans crowning it the top vector database and pragmatists saying to stick with Postgres + pgvector unless your scale or features demand more.
Milvus just blasted past 40,000 GitHub stars, and the comments instantly turned into a Postgres vs Milvus cage match. One camp is waving the “just use Postgres” meme, with pietz asking if the old rule still stands: pgvector is cheaper, familiar, and fast enough. The other side fires back with big–league vibes: LuminaWang7 declares Milvus “the most popular vector database in the world,” while notachatbot123 keeps it simple—Milvus is open-source and built for “GenAI,” meaning apps that use AI to find things by meaning, not just exact words.
Behind the noise, there’s real news: the team rebuilt Milvus twice and in 2025 shipped native hybrid search—mixing keyword search with meaning search in one engine—so teams don’t juggle an extra Elasticsearch cluster. Fans say that’s fewer pieces, fewer headaches. Skeptics clap back: Postgres already does enough, and stars aren’t production revenue. The thread injected classic web jokes—“Postgres can do your taxes,” “stars ≠ scale”—plus calls for head‑to‑head benchmarks and cost breakdowns. However this shakes out, Milvus is now the lightning rod: builders want speed and simplicity, and the comment section brought both the cheers and the chair‑throwing. Expect more benchmarks, receipts, and spicy replies in the days ahead from both camps.
Key Points
- •Milvus surpassed 40,000 GitHub stars in 2025 after reaching 35,000 in June 2025.
- •Milvus’s development began in 2017, with open-sourcing of Milvus 0.10 in 2019 and key milestones in 2020–2021, including joining LF AI & Data, shipping 1.0, graduating, and winning the BigANN challenge.
- •Enterprise requirements led to a ground-up rebuild, and Milvus 2.0 introduced a decoupled, cloud-native architecture in 2022.
- •In 2024–2025, Zilliz was recognized by Forrester, and Milvus saw broad production adoption across multiple AI use cases and industries.
- •Milvus 2.5 (2025) added native hybrid search, unifying full-text and vector retrieval to reduce reliance on separate Elasticsearch/OpenSearch deployments.